> On Thu, Jul 07, 2011 at 09:21:22PM +0100, Matthias Scheler wrote: > > > We have nearly five hundred test files. > > > > This is a problem why? > > Because most of these are potentially a subject of similar hypothesis. > > > > ... whereas the libc-tests should be strictly about functional > > > verification. > > > > This about functional testing. This is whether the environment code works > > with and without real locking, with and without "libpthread" present. > > I'd say this is about non-functional testing, but this is just semantics. > The kind of granularity that is seeked here does not give that much useful > additional information. Alternatively, by linking the old "t_environment" > test with pthreads, we now know also whether this is caused by linking alone.
considering that we've had bugs related to this in the past i think that having both types of this particular test is entirely the right thing to do here. it may be a special case in that we don't want to have a slew of _pthread versions, but IMO this one should be put back. .mrg.
